AI Paper Atlas

Research map / Label-efficient and robust learning / Test-time adaptation

Learning to Receive Help: Intervention-Aware Concept Embedding Models

Mateo Espinosa Zarlenga, Katherine M. Collins, Krishnamurthy Dj Dvijotham, Adrian Weller, Zohreh Shams, Mateja Jamnik

NeurIPS 2023 · Label-efficient and robust learning › Test-time adaptation · approach: Extrapolation and single-sample adaptation · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
IntCEM loss (lambda_roll=1) vs standard CEM trainingCEM · MNIST-Addtask AUC: 90.13 ± 0.41 -> 92.07 ± 0.20Improved
IntCEM (lambda_roll=0.1) vs CEMCEM · MNIST-Add-Incomptask AUC: 86.61 ± 0.75 -> 89.63 ± 0.13Improved
IntCEM (lambda_roll=1) vs CEMCEM · CUBtask accuracy: 79.00 ± 0.61 -> 78.03 ± 0.58Hurt
IntCEM (lambda_roll=1) vs CEMCEM · CUB-Incomptask accuracy: 75.52 ± 0.37 -> 74.70 ± 0.22No clear effect
IntCEM (lambda_roll=5) vs CEMCEM · CelebAtask accuracy: 31.04 ± 0.96 -> 38.09 ± 0.26Improved
Lower lambda_roll (5 -> 0.1) in IntCEM (ablation)IntCEM · CUBmean concept AUC: 89.63 ± 0.42 -> 93.39 ± 0.22Improved
IntCEM (lambda_roll=5) vs CEMCEM · CelebAmean concept AUC: 85.28 ± 0.18 -> 88.25 ± 0.24Improved
Joint CBM-Logit vs CEM on concept predictionCEM · MNIST-Addmean concept AUC: 85.51 ± 0.26 -> 89.21 ± 0.71Hurt
Random test-time interventions on IntCEM vs baselinesIntCEM · CUBtask accuracy: ~10% with ~25% of concepts intervenedImproved
Test-time policy: learned psi vs CooP query timeIntCEM · CUBpolicy query time (ms): 4.77 ± 0.10 -> 2.90 ± 0.20Improved
Learn psi end-to-end instead of post-hoc behavioural cloning (BC-Skyline) (ablation)IntCEM · CUBtask accuracy vs interventionsImproved
Remove learned train-time policy psi (random trajectories, IntCEM no psi) (ablation)IntCEM · CUBtask accuracy vs interventionsHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy